Routing Optimization By Concurrent GeneticAlgorithms
Alessandro Lo Genco, S. Lopes, Giuseppe Lo Re, M. Tartamella · WIT transactions on information and communication technologies · 1970
A parallel implementation of genetic algorithms is applied to routing protocols with low bandwidth consumption. In particular, the paper discusses the (LSP) link state packet protocols. The first part of the paper deals with network topology and transmission parameters, together with the structure for storing the network. The second part discusses the Genetic Algorithm implementation. To this end, it considers the generation of the initial population that is a subset of all the possible paths connecting couples of nodes. As far as the mating and mutation policy is concerned, a strategy is proposed that allows the algorithm to replace segments of the entire path The implementation is carried out in parallel, thus letting different populations to evolve separately. Subsets of different populations migrate periodically to avoid the populations to persist in some local minima. These are the equilibrium states, where no better path, with a lower cost, can be found for a given period. As for conclusions, comparisons between the results of the sequential and distributed implementations of Genetic Algorithms are reported.